Question: Torsion PCA

[[e-mail]]

Dear Bio3D Team,

I have a question about Torsion PCA.

I would understand for Torsion PCA, first Cartesian coordinates are converted to internal coordinates and then variance-covariance matrix is calculated.

I would appreciate your response.

Mamta

[Xinqiu Yao]

Sorry, I don't understand your question. Please provide more detail about what you expect to calculate and what the problem is, as instructed in the issue tracker.

[Xinqiu Yao]

Q&A, not a bug.

[[e-mail]]

Dear Xin,

I would like to calculate dPCA for my Trajectory. Bio3D has Torsion xyz and Torsion PCA calls.

My understanding is for dPCA call is, from a MD trajectory, first Cartesian coordinates of an atoms are translated to their internal coordinates and then PCA is calculated.

In Bio3D for my trajectory which has xyz coordinats of atoms , if I use torsion.xyz call it will give me a matrix of torsion angles and pca.tor will calculate PCA from that matrix.

This I think is different from dPCA calculation.

I would appreciate if your can describe how Torsion PCA is calculated in case I understand it otherwise.

Mamta

[Xinqiu Yao]

Hi,

Could you explain a little bit what you mean "dPCA", or give a link to proper references. I never heard of it, so don't know what exactly you want to calculate.

[[e-mail]]

Dear Xin,

This paper describes dPCA: Dihedral angle principal component analysis of molecular dynamics simulations. Altis A1, Nguyen PH, Hegger R, Stock G.

Please find the link below: https://www.ncbi.nlm.nih.gov/pubmed/17614541

Mamta

[Xinqiu Yao]

Hi Mamta,

Thanks for providing the link to dPCA. It seems an interesting method, but I don't think it is the same as what pca.tor() does in bio3d. In pca.tor(), angles are not converted to "cosines" but instead directly fed into PCA.

You are welcome to contribute your own codes of dPCA. If you are willing to do so, please test your codes and let us know when you are ready to submit.

[[e-mail]]

Dear Xin,

Thank you for your response.

Mamta